It's pretty obvious why AMD is able to scale x86_64 faster than anyone else... they invented it. Intel took it, implemented it in broken ways to win benchmark comparisons and win IT buying managers' hearts and minds, locking everyone into their chipsets. Finally, all those broken implementations were exposed, and AMD finally had the capital to do things right at scale.<p>RISC or MIPS or any other CPU in that "minimized" class doesn't really have the room to scale as fast.
I remember in 2006 when Core processors were launched - when AMD was the top leader of desktop processors. There was a huge fanbase rooting for Intel to beat AMD in some sort of a competitive fanboy arena. Conroe architecture and the introduction of tick-tock methodology from Intel was absolutely incredible and totally destroyed the competition.<p>The tides have turned, so perhaps I should now start preparing to root for Intel :)
It's about darn time.<p>GPUs have been giving us <i>so</i> much more bang for the buck than cpus - the latest nvidia gpu is 54 billion transistors @ 826mm2<p>I know gpus are not the same as cpus, but I think amd's experience there is getting us cpus with comparable transistor counts at comparable economics.<p>intel doesn't seem to publish numbers, but I think the top $13,000 28-core intel cpu is 8b transistors<p><a href="https://en.wikipedia.org/wiki/Transistor_count" rel="nofollow">https://en.wikipedia.org/wiki/Transistor_count</a>